- From: Chris Lilley <chris@w3.org>
- Date: Sat, 1 May 2004 00:26:20 +0200
- To: "Richard Ishida" <ishida@w3.org>
- Cc: www-international@w3.org, "'Deborah Cawkwell'" <deborah.cawkwell@bbc.co.uk>
On Friday, April 30, 2004, 8:06:18 PM, Richard wrote: RI> Can anyone think of a general way of testing whether the RI> language declared in an html tag is recognized by a user agent - RI> that we can use in a test suite? There are various ways it can be used, is the trouble. Its sort of metadata. An editor can use xml:lang to assist with spell checking, for example, and I believe Amaya does this (try a mixed French and English document. But a user agent is not required to support that. I recall François demonstrating lang for selection of appropriate glyphs on a mixed Chinese and Japanese Unicode document. But a user agent is not required to support that, either. Actually the definition of an HTML user agent doesn't really exist. RI> This would be particularly useful to test, for example, RI> whether the langauge declaration in a meta statement has any RI> impact on the document - my assumption is not, but I'd like to be RI> sure. RI> The difficulty is finding a test that doesn't rely on some other thing working. I agree. RI> We could use a :lang selector to test for recognition, except RI> that the test would fail on IE because :lang doesn't work. Last I looked, if something doesn't work in a particular implementation that didn't mean that the spec was broken. RI> We could maybe check whether fonts are applied in CJK (see RI> for example [1]), but we can't guarrantee that that behaviour is RI> widespread either. Right. RI> Maybe a JavaScript routine could look at the DOM? Well it could. RI> Any ideas? SVG has a notion of language-specific glyphs in SVG fonts. But that, as you say, depends on a whole set of other things working. -- Chris Lilley mailto:chris@w3.org Chair, W3C SVG Working Group Member, W3C Technical Architecture Group
Received on Friday, 30 April 2004 18:33:22 UTC